gsaWilcoxSurv: Gene Set Analysis GSA, experimental implementation
In bernau/survHDExtra: Addtitional Features for package survHD

Description Usage Arguments Details Value Author(s) See Also Examples

Test whether a set of genes is highly ranked relative to other genes in terms of a given statistic, for example a ranking based on univariate Cox coefficients. Genes are assumed to be independent.

## S4 method for signature 'gsagenesets,missing,missing,character,numeric'
gsaWilcoxSurv(
gmt, genenames, statistics,  p.value=0.1, cluster=FALSE,
cluster.threshold=0.3,...)
## S4 method for signature 'gsagenesets,ExpressionSet,Surv,missing,missing'
gsaWilcoxSurv(gmt,
X, y, genenames, statistics, ...)

`gmt`	An object of class `gsagenesets` or a list of gene sets.
`X`	An object of class `ExpressionSet` or matrix.
`y`	An object of class `Surv`.
`genenames`	Alternatively to `X` and `y`, a `character` vector of gene names rcorresponding to genes in `gmt`.
`statistics`	`numeric` vector for genes specified in `genenames`, any genewise statistic by which genes can be ranked.
`p.value`	Only report gene sets with p-value lower than this cutoff.
`cluster`	If true, then significant gene sets are clustered and ranked together.
`cluster.threshold`	Minimum overlap of gene set clustering.
`...`	Additional arguments passed from alternative S4 signatures or to the wilcoxGST function.

A convenient wrapper around the wilcoxGST function from the limma package. See the limma package documentation for details.

This method makes it a little bit easier to test gene sets from a GMT file with the wilcoxGST function. It further adds a clustering of gene sets that passed the p.value threshold. The cluster parameter cluster.threshold specifies the minimum overlap of genes (default is 0.3 or 30 percent) in the clustering.

An object of class gsaresults.

Markus Riester markus@jimmy.harvard.edu, Levi Waldron lwaldron@hsph.harvard.edu, Christoph Bernau bernau@ibe.med.uni-muenchen.de

gsaReadGmt, gsaTranslateGmt

    library(survHD)
     library(survHDExtra)
    set.seed(100)
    # create some random data
    x<-matrix(rnorm(1000*20),ncol=20)
    dd<-sample(1:1000,size=100)
    u<-matrix(2*rnorm(100),ncol=10,nrow=100)
    x[dd,11:20]<-x[dd,11:20]+u
    y<-Surv(c(rnorm(10)+1,rnorm(10)+2), rep(TRUE, 20))
    genenames=paste("g",1:1000,sep="")
    rownames(x) = genenames

    # create some random gene sets
    genesets=vector("list",50)
    for(i in 1:50){
        genesets[[i]]=paste("g",sample(1:1000,size=30),sep="")
    }
    geneset.names=paste("set",as.character(1:50),sep="")
    gmt <- new("gsagenesets", genesets=genesets, geneset.names=geneset.names)
    gsa.res <- gsaWilcoxSurv(gmt, X=x,y=y,cluster=FALSE,p.value=0.3 )

    # show the similarity of significant gene sets
    plot(gsa.res)

    # display a barcode of up to two gene sets.
    plot(gsa.res, type="barcode",geneset.id1="set22", geneset.id2="set33")
    
    library(genefilter)
    # use a pre-ranking of genes
    genes.ttest = rowttests(x, as.factor(c(rep(1,10),rep(2,10))))
    gsa.res.tt  <- gsaWilcoxSurv(gmt, genenames=rownames(x),
        statistics=genes.ttest[,1])

    # now test some gene signatures on our Affymetrix example data (Beer et. al
    # 2002)
    data(beer.exprs)
    data(beer.survival)
    library(hu6800.db)
    library(annotate)

    gmt <- gsaReadGmt(system.file("extdata/ovarian_gene_signatures.gmt", package = "survHD"))
    # the Gmt file contains gene symbols, so we translate it to Affymetrix 
    genes <- getSYMBOL(rownames(beer.exprs), "hu6800")
    gmt.affy <- gsaTranslateGmt(gmt, beer.exprs, genes)
    gsa.res <- gsaWilcoxSurv(gmt.affy, beer.exprs, Surv(beer.survival[,2], beer.survival[,1]))